Pristine Audio Favourites

[Gordon]

In the discussion above frankwm raised the question of the quality of conversion/restoration of Klemperer's Brahms' 4th symphony. His version is posted at his site - link given in Alain M's post above. I have listened to it and also the Pristine download I already had.

Here are some waveforms of the first movement in both linear and log plots. You can see some differences. Also plotted below are some spectra of a 1 minute section around 2 minutes in; the top one is frank's.

I have some comments to make on the plots here and also on my listening session. Those to follow later.

[frankwm]

Bear in mind that I stated this was an initial/experimental transfer which had some 'suspension' defect - apparently affecting the RH waveform trace (it's a JVC MM with beryllium cantilever/shibata stylus).

The LF resonance should be around 8Hz - not 15Hz..as for an 'identical' one (JVC 'X-1' Laboratory series) used for many of the 'Historical' issues: but whilst there was no elastomer degradation in that - two others (the above was 'NOS') have faults.
I have no particular need to use them - but the JVC design was about the finest/most dynamic MM cartridge of that period - and was designed for CD-4 LP's (hence the frequency-range)

The graph on the link differently derives from a Shure M95HE (NOS stylus) - rarely used - but discloses the channel imbalance between original/later masterings.

The Rose shows the fizzed-up treble (JVC's are treble accurate/flat to above 20kHz, with a minor <1.2dB trough 2-10Khz - as usual with most MM cartridges (which I sometimes correct, depending on the LP's transfer characteristics)

'Exotica' - such as the Dynavector Diamond cantilevered are 'ruler-flat' - which I used for the Brahms 1/Furtwangler (also linked to Rose's effort: same LP' transfers used there...) and the Souzay 1956 Decca recital (both on that blog) : however you are not seeing the same transfer in the Klemperer examples....but the original 1960 LP on my files is sonically most superior to the later mastering in the SLS box I have (and I know the sound of this LP very well as I originally bought a late copy as a deletion from 'Orchesographie', in November 1971 - with '-1' being used for side two (the 'spare' 1960 mastering)- with a transistor recut for side one: Roses are from transistor recuts in the SLS box-set.

The issue of 'phase anomalies' on Rose's Klemperer needs to be addressed; also listen to just the output from the RH channel of his; Weird, eh..?

This will leave FF bemused..and I hadn't intended responding again...but, if you are going to 'analyse' this here (instead of on my blog)...

[Gordon]

Thanks frank, good stuff. I'm sure others will be interested in some of this too so discussion here useful.

Quick comment now because today is wife's birthday and this is a bit of truancy just beofre we go out!! Will comment on listening later but quick one now is that it is clear that the AR transfer has lifted the treble and there is a bit of fizz in the RH channel. There are also dynamic differences which show up on level meters but are not so obvious when listening. There's not much in it most of the time but it's there. The running times are exactly the same [and the same as the CD] which is a good sign that people's turntables are running true.

Here's some detals of the issue of the recording:

Begun at Kingsway Hall November 1st 1956 and completed March 28/9th 1957. Probably recorded on modified BTR2. BTR3 [REDD38] did not enter service until late 1958 along with a new desk REDD37. The .wav file sizes on ripping the 2 CDs are exactly the same.

Issued first on Columbia SAX 2350 [S] and 33CX1591 [M]. Later reissued as ASD 2708 [EMI/HMV label now] and in a box set SLS 804 and then SXLP 30214. I can find no CD issue on 747 xxx 2 series but on this CDM 769 649 2 DRM Angel EMI Studio series pressed by EMI in 1989 at Uden in Holland - 769649 @ 1 and 1-1-6-NL. Same catalogue no issued in a box of Romantic Symphonies by Klemperer in 1994 but pressed in W Germany by Sonopress: C-9323/CDM769649 2 A in 1994.

These further plots are taken from the EMI/Sonopress CD version which should not suffer vinyl playback issues and perhaps reflect the master tape. The CDs are identical. We need to be wary because the 2 tapes used by the LP cutting engineers [1958/9 and ca1971] will have been EQ'd for that purpose [I doubt they cut on the fly!] whereas the CD MAY have come from a different EQ or even the original. Your comment re transistor cutting amps is a good one [Neumann?]. I think EMI would have been still on valves in 1958/60 for the first cutting of SAX 2262/2350 but moved on by ASD 2708 recut and then the Box. I suspect that the SXLP, being a re-issue, might have used the old metals? One would need to look at the disc. All up the CD sourcing is perhaps more direct and so less obscure than the various vinyls. At least you know what you did.

Here's the plots: waveforms and then spectrum. The CD spectrum is closer to you [order top - bottom is You, CD, AR] than AR but waveforms show dfferences. Tape hiss on CD more prominent than AR too and you have left in some surface noise. The surprise is how little there is around 12k and above.

[frankwm]

'Leaving surface noise in' (I've made note in the comment's re that) really relates to the fact that Rose uses 'Izotope' de-click/de-crackle software; and I can assure you (as can also use it..) that even @ the near minimum setting which I've used (@'1.4 relative to the '1' start / '10-end') for 3x PYE-pressed LP's on that blog, that transient information/fine detail-resolution is being damaged/compromised at my chosen level (and 'surface-noise' or 'easily removed crackle' - sic) won't be eradicated until much higher intervention levels are employed; so the 'jacked-up' treble prevalent on Rose LP transfers is an effort to disguise information loss (which can be considerable: whether you can hear it or not - I can; especially on the computer speakers.....

Rarely will you find mention of that in Reviews: one such is by a former R3MB member in his comments re: Elgar: Symphony 2/Barbiroll: 1954:

http://www.musicweb-international.com/classrev/2012/Oct12/Elgar_sy2_PASC337.htm

- but no LP source is stated/claimed: however the LP doesn't have 'glassy strings' - and there is noticeable loss of woodwind detail: which would match the above comment about LP's.

EMI used 'Westrex' cutters - with a limited response for stereo LP's above 8kHz until about the mid-60's. The lack of HF over 12kHz possibly is caused by the Kingsway Hall acoustics (in terms of 'stray' harmonics being captured?)

If you download my Hoffnung 1956 concert, on the main MP blog, you will see the 1973 stereo LP transfer goes up to the 22kHz limit of the CD-RW dub (Arnold's Grand Grand overture on transient cymbal crashes (as do the early Decca mono (BTR 2 tape-decks @30ips used until 1953 -then Ampex @15ips] recordings: though possibly still used for subsequent transfers of those - and go up to 30kHz on the 24/96 files).

Later SXLP's got additional re-cuts: hence my comments re: Brahms Symphony 1.

That's enough for now: but you should be hearing considerable differences between the Brahms 4 'versions' as 'noise-reduction software' can leave the graphs virtually indistinguishable in appearance - but the 'transient/fine-detail' has gone/instrumental 'smearing' then also occuring.

[Gordon]

Thanks frank, see comments and some listening notes below:

'Leaving surface noise in' (I've made note in the comment's re that) really relates to the fact that Rose uses 'Izotope' de-click/de-crackle software; and I can assure you (as can also use it..) that even @ the near minimum setting which I've used (@'1.4 relative to the '1' start / '10-end') for 3x PYE-pressed LP's on that blog, that transient information/fine detail-resolution is being damaged/compromised at my chosen level (and 'surface-noise' or 'easily removed crackle' - sic) won't be eradicated until much higher intervention levels are employed; so the 'jacked-up' treble prevalent on Rose LP transfers is an effort to disguise information loss (which can be considerable: whether you can hear it or not - I can; especially on the computer speakers.....

Understood. I have no experience of that software so cannot comment. However I can appreciate that if your ears are highly tuned from listening to a lot of vinyl using a variety of different cartridges etc you will be more sensitive to sound quality than someone that listens more casually.

Rarely will you find mention of that in Reviews: one such is by a former R3MB member in his comments re: Elgar: Symphony 2/Barbiroll: 1954:

http://www.musicweb-international.com/classrev/2012/Oct12/Elgar_sy2_PASC337.htm

- but no LP source is stated/claimed: however the LP doesn't have 'glassy strings' - and there is noticeable loss of woodwind detail: which would match the above comment about LP's.

Is it possible that this glassy string quality is the result not necessarily of the processing applied but the cartridge/arm etc used as well? You have that experience of knowing what differences arise in different pick-ups. Even without discussing digital processing people do get exercised about their favourite purely analogue preferences.

EMI used 'Westrex' cutters - with a limited response for stereo LP's above 8kHz until about the mid-60's. The lack of HF over 12kHz possibly is caused by the Kingsway Hall acoustics (in terms of 'stray' harmonics being captured?)

Not sure about that for lack of detailed knowledge. KH was prized for its acoustic of course especially for large scale works so it would be a surprise if it contributed to loss of HF. I know the weather affected the sound because Robert Gooch [Klemperer Bruckner among others] told me that his experience was that too many winter coats draped here and there didn't do any good!! Chris Parker also was of the view that that Hall was better in the summer when it was warmer and less damp. The place was not heated well and suffered all sorts of neglect. EMI and Decca refused to get the floor revarnished though for fear of the effect on the sound!! At one time they wanted to fill the basement with sand to deaden the LF a bit and to keep out the worsening tube noise.

If you download my Hoffnung 1956 concert, on the main MP blog, you will see the 1973 stereo LP transfer goes up to the 22kHz limit of the CD-RW dub (Arnold's Grand Grand overture on transient cymbal crashes (as do the early Decca mono (BTR 2 tape-decks @30ips used until 1953 -then Ampex @15ips] recordings: though possibly still used for subsequent transfers of those - and go up to 30kHz on the 24/96 files).

Done that thanks. I had the LPs once but got rid of them and always meant to get the CDs but never did. I see what you mean about the spectrum!! Wow! I would not have believed that of 1956. I have to say though that there is a slightly roughish edge to the sound throughout. What mics, probably, as you say, a crossed pair - it was usual to see one during concerts in the 70s. I wonder of they tamed to the often toppy sound of capacitor mics, a noticeablelift above 14kHz.

Later SXLP's got additional re-cuts: hence my comments re: Brahms Symphony 1.

That's enough for now: but you should be hearing considerable differences between the Brahms 4 'versions' as 'noise-reduction software' can leave the graphs virtually indistinguishable in appearance - but the 'transient/fine-detail' has gone/instrumental 'smearing' then also occuring.

Yes, perhaps we are doing this subject to death!! I don't regret that because underlying all the measured responses etc and the factors affecting listening responses is personal choice and the gift of acute hearing or at least well trained hearing. My experience of working with golden eared people was that they were not infallible. As I said before the variables are such that definitive conclusions are all but impossible without tight control over them.

You have used and emphasised the word "considerable". Perhaps that is where some of the differences of opinion might lie. We might hear the same difference but make more of it? A question not a statement. As you will see from my listening notes below, I do hear those things you speak of but I would not use the word "considerable" to describe them, noticeable, yes; there is more than one factor at play in assessing a preference too an so it seems to me that compromise is inevitable. You might say there is no room for compromise - but audio is riddled with compromise and the "best" is an optimum given certain conditions. When we lived in London we used to go to the RFH regularly when they ran those subscription series concerts. I always came home to realise what a waste of time HiFi was and as a substitute it wasn't worth getting into a sweat about.

This post is another long one but I promise it is the last such!! Honest

Perhaps there is room for a new thread about the business of restoration. When I worked in R&D and we were faced with some new technical issues I would always go back to first principles. In this case what is restoration for? Why do we do it?

Listening Notes, Brahms 4:

I listened to all 3: Pristine Audio [Andrew Rose – AR], Pristine Classics [PC] and the EMI CD. I only used the 1st and3rd movements. Otherwise I’d still be at it. I used a DAC into the Cyrus P/A for the digital files and direct CD so that I could switch between CD and Files having got the levels as near equal as possible – the CD is cut lower so that the ripped wav file plays back quieter by about 2dB. That was for listening on the main system in the lounge. I then used headphones from the computer using all wav files and a playback package to enable me to switch between them all at will. The CD was 16 bits of course.

Listening via the main system I could hear little difference between them all at first but then as I “learned” I could discern some small differences consistent with the HF lift in AR compared to PC and CD. The CD and PC also had slightly more air and the AR a slightly veiled quality. I would not say there was much in it, there were differences but I would not describe them as massive. I could not swear that the HF Fizz mentioned before in the RH was that audible and I could not really say that the stereo image was that different despite the comments made on the PC blog. However in the 3rd movement that triangle at 1.27 and again at 6.10 is revealing. In all 3 the earlier softer sound is comparable but the AR seems to have a problem with it in the louder later case where the triangle has a backdrop of loudish orchestral competition. The AR causes the triangle to stand out more but it seems quite a harsh sound. In the CD and PC it is not quite buried but not prominent either – that HF lift again perhaps.

The surface noise audible in the PC was OK but I have an aversion to it, always have, one of the several reasons I stopped buying vinyl way back. The CD tape hiss is also audible [it has been through some kind of DRM vintage 1989 so expect that it has been reduced but by how much we don’t know] but more acceptable as a reminder of the carriage medium. Tape hiss in AR is much reduced implying heavier intrusion.

After this, given the choice I would choose to listen to the CD in preference to either LP transfer preferring to compromise any slight sound degradation of the DRM and CD for the quiet unobtrusive tape hiss and absence of vinyl characteristics and recovery issues ie cartridge set up etc. The variables in vinyl playback are considerable. Surface noise apart the CD and PC were closer in sound quality than the AR. You will not agree with my conclusion I am sure, but I have reported my experiences. If I have "bad ears" well too bad, that's all I have. It doesn;t stop me enjoying my music.

Listening on headphones opened up more of the differences when the HF lift was more apparent in AR. Those triangle events were accentuated. Listening to the sound itself – ie forgetting the surface noise there is a veiled quality to the AR – I don’t think it’s massive but it is plainly audible. On balance I could settle for either if I just wanted to hear this performance - but would rather there was less surface noise - it’s just my preference - so would compromise [sorry] on the AR if pushed. Headphone listening confirmed that I would rather the CD to both.

The two transfers illustrate well the dilemma facing a restorer. Go for a commercial approach with quiet backgrounds at the expense of some of the finer elements of the sound to satisfy a market that demands that there should be little evidence of clicks or surface noise, or to retain as much of the music as the recovery system [favoured cartridge etc] will permit and just remove the worst of the clicks etc. The latter is closer to a vinyl experience ie the digital file is a nearer clone of the vinyl. Pay your money [or not in the case of PC] and take your choice.

Tell you what though, this sort of intensive listening doesn’t half take time.

[frankwm]

Generally your conclusions are OK to Professor Tin Ear.. who'd only remark that some will spend £100,000+ just to replay the Klemperer 1960 vinyl... and anyway suspects people give those transfers 'an unnecessary dose' of noise-reduction - even when already noise-reduced (as with Weingartner's 1927 Beethoven 5/6 from shellac; which you can compare to Obert-Thorn's effort via the sample, linked)... despite saying those transfers were 'fantastic'....

As an experiment, you might try adding the treble boost (AR's seem usually around +5-6 dB: ie +50% on old valve-era Quad/Leak knobs; and nearly 'full blast' on my upper-class Jap gear) to mine/EMI CD to hear what the effect is: I would expect it would be quite unacceptable; but if you'd reduced the inherent resolution (AR), then.....

Incidentally, there's an updated comment on 'Capstan' software (used by AR) @ ARSC:

http://listserv.loc.gov/cgi-bin/wa?A2=ind1312&L=arsclist&T=0&O=D&P=28974

Re: Mahler 9/Walter-VPO: that was issued 1973 as WRC SH193/4 (ie, after the Kajanus/Sibelius double-album;- included on my blogs, with all the other pre-war 'Society' albums, etc) and was transferred by A.C.Griffith.
If Hardwick, who joined EMI c.1979, is credited, then perhaps he'd 're-mastered' those tapes.
NB: EMI will claim copyright-protection on that CD; so it's 'unlawful' to clone/sell commercially....

[Gordon]

The problem with "intrusive intervention" is it isn't easily correctable and the more you do the worse that gets. As you say, the old tone control knobs could be used to deal with simple EQ differences but are next to useless for anything complex. Modern amps don't give you any anyway!! I still keep an old QUAD 44 P/A [don't laugh] hooked into the system through which I play my few remaining vinyls on the Linn/Ittok/Ortofon MC25 having tweaked the MC card to match - MCs are easier than MMs in this respect. I did intend to get a good phono outrigger but baulked at how much a good one would cost for little use.

I don't understand how that Capstan thing works, must do some research and look at that link, for which thanks.

BTW the latest Pristine Audio newsletter came in a few minute ago with this attached, bold in original, looks like there was some flak - with a K:

24-bit article: reaction

My article last week proved controversial in some quarters, and I have to admit I simplified one aspect of my descriptions for the sake of making the maths easier to convey (i.e. my description of a binary zero of 16 zeros is actually fully minus on a waveform - the bottom of a wave which is actually very loud, and not in fact silent! - but I felt it simpler to explain things the way I did than get into the intricacies of mapping positives and negative decimal sample values onto a positive-only binary scale).

One point I did make and would like to briefly return to is the psychology of all this. When I prefer to listen to one of my 24-bit recordings at home it's because I know that's the master copy, and I know it must be better, even if I don't hear a difference between this and its 16-bit counterpart (carefully prepared by me!). The knowledge that something may be missing, even if you aren't sure you can hear what's missing or whether it's relevant, is a powerful incentive in itself to upgrade to 24-bit sound.

A fascinating, if brief comment on our Facebook site by the Grammy Award-winning classical music producer Michael Fine seems to illustrate this point rather well:

I was intrigued by your cogent article on the science of bit rates. Of course, the moment, our perceptions, and pre-conceived notions play a huge role in how we listen. I produced a recording of Mahler 2 and sent the label a 44.1/16 bit CD master. In Japan, in addition to a standard CD, it has recently been released as a rather expensive SHM CD. Several of my highly respected Facebook friends have been extolling the new re-mastering but of course, this edition was made from the same master as the standard release. But they insist it is substantially better. There is also the classic case of the poor critic who reviewed six budget recordings of a recording I produced in eastern Europe in the early days of the CD. It was pirated and appeared on several labels with the names of conductor and orchestra changed. The hapless writer compared the same recording in six different incarnations and detected substantial differences between them!